Last week, Dr Andrew Ng came to Singapore and I have the privilege to attend one of his session. His talk was very insightful, and fantastic! Because he did not mentioned much on Generative AI (GAI) but the whole landscape right now after the advent and progress of GAI rather. (Yes, I find it a bit too much these days on GAI, cos as usual, with a snap of the finger, many GAI “experts” were born.)
No, this newsletter is not going to be about GAI, but rather as the title goes, I wanted to share my thoughts on a comment made by Dr Andrew Ng rather. Dr Ng mentioned that the growth in value Reinforcement Learning (RL) is going to bring will not increase that much in the next 3 years. To this I totally agree for the following reasons:
I hope to keep the newsletter open to as many as possible. If you have benefited much from the newsletter and will like to join my cause, consider making a “book" donation. Link at the bottom of the newsletter or here. :)
The barrier to adopt RL is higher than GAI right now. If you do not need “customisation”, GAI can be easily adopted with good prompt engineering skills. For RL, there are assumption that the companies have hit a certain maturity level in adopting Machine Learning, and that will be 0.01% (guess-timate) of the companies out there. The learning journey will be long for most companies even with management support, and that is a big IF in most organizations I have seen so far.
Talents in RL is less and few. How so? Because RL requires very strong mathematical background followed by experience in framing the different use cases. Framing what? Constraints and objectives, followed by converting them into mathematical symbols for manipulation and computation.
Last but not least, GAI will be sucking up most of the research dollars out there given its deployment to commercial cycle is very short.
The point is the growth of RL will be slowed down because of GAI, and only the larger companies and research universities will still be able to work on it. However, I am of the belief that RL will be the key ingredient in building smarter machines.
What are your thoughts? Please share them in the comments below! :)
Recommended Content
For subscribers who are not familiar with Reinforcement Learning, do check out wiki here. <Wiki>
If you are looking for a course on Reinforcement Learning, check out this YouTube series by David Silver <YouTube>